Tagged: Bioscience

Polar bear attack paper invalidated by non-independent analysis

disclaimer

From Bart Verheggen’s blog, where he clearly identifies himself as a co-author of a paper in a post promoting it

Stephen Lewandowsky has co-authored (yet another) paper attacking climate skeptics. His colleagues-in-arms this time are long-time climate consensusite Jeff Harvey , Bart Vergheggen, and a cohort of ecologists along with Michael Mann. First author Harvey is well-known to climate commenters as a rant-prone passionate bulldog for the climate cause.

The main supposed finding of the paper is that zoologist Susan Crockford is the source of a number of skeptical blog posts. Harvey and colleagues claim a large figure (80%). The authors then claim to identify a ‘majority-view’ position in the polar bear literature, which they say is diametrically opposite of the Crockford-based blog position/s.

Polar bear alarmism has a checquered history and scientists Ian Stirling, Steven Amstrup and Andrew Derocher have been prominent proponents. All three have made several statements pushing a specific line – that polar bears are under severe threat, that anthropogenic global warming is the cause, and that their ability to adapt to changing conditions is limited. Of note here, the paper is co-authored by Ian Stirling and Steven Amstrup. Susan Crockford has been critical of both scientists on her blog and other venues.

My first thought was on seeing the Harvey et al text was whether the so-called ‘majority-view’ papers mainly cited Stirling, Amstrup and Derocher papers in support of their views. Did they identify a view present in the literature which traced its antecedents to their own papers?

It turns out the situation is much worse.

Of the 92 papers included in the study,  6 are labeled ‘controversial.’ Of the remaining 86, 60 are authored or co-authored by Stirling or Amstrup, or Derocher. That is, close to 70% (69.76%) of the so-called ‘majority-view’ papers are from just three people, 2 of whom wrote the attack paper themselves.

papers

Papers analysed by Harvey et al. Hightlighted in yellow – papers co-authored by Stirling, Amstrup or Derocher.

In other words, Stirling and Amstrup did not discern an organically coalesced body of opinion from several polar bear papers by sifting through the literature. They did not even uncover a body of literature supporting a particular stance that cited their own work, as self-referential as that might have been. They ‘found’ their own papers to constitue a ‘majority-view’ in the polar bear literature!

Stirling and Amstrup attack Susan Crockford for not following the ‘majority-view’ and the ‘majority-view’ is what’s expressed in their own papers.

But there’s worse to come. The authors list 6 papers as being ‘controversial’ for  eliciting ‘critical comments and discussion in the peer-reviewed literature.’  It turns out Stirling, Amstrup and Derocher themselves wrote comments to 4 out of 6 of these papers. Put another way, Stirling and Amstrup labeled papers they did not like ‘controversial.’

Quartz %d

It is no wonder the ‘majority-view’ (green triangles above) displays such a tight cluster of perspectival homogeneity. It is not a majority view but rather a minority one, of just three scientists. The near-absolute lack of variability in opinion along the PC1 axis is likely just due to standard boilerplate alarmist text in the papers of Stirling, Amstrup and Derocher, repeating the mantra of polar bear doom from melting ice, rather than any emergent phenomenon in polar bear literature.

A true majority view (if there can be such a thing) can be discerned only if a representative sampling of the polar bear literature is carefully assessed, with attention to their scientific content (as opposed to mere headcount), the nature and strength of supporting evidence presented and the caveats that scientists are careful enough to always include. In such a setting opposing viewpoints cannot be dismissed as being controversial merely because they oppose one’s own views.

The paper has several hallmark characteristics of a Lewandowsky piece: the language is dominated by ad hominen attack (for e.g, the word denier occurs 31 times) and the text is notable for a number of false statements. The authors purport to analyse ‘the views’ of blogs but ascribe views to the blogs themselves followed by analysis of the same views. Last but not the least, the full data from the paper is not made available. But the fatal flaw of non-independent analysis by the paper’s authors renders its conclusions invalid.

Advertisements